From: Ben Goertzel (ben@goertzel.org)
Date: Sun Jan 06 2002 - 12:14:39 MST
> > For an AI at this phase of development, the Open Mind /
> mindpixel kind of
> > data could potentially be informative.
>
> It would be a very brief stage of development - one is assuming that the
> AI has reached the point of general intelligence and language
> understanding, but not yet the point where it can freely draw on the much
> larger corpus of information represented by the Internet.
And just *how brief* this stage is going to be, we really don't know, do we?
You assume it will be .04 picoseconds, so that OpenMind is useless. (Yes,
it's
my turn to polemically exaggerate.). Pushpinder assumes it will be years,
maybe
decades.... I of course, as the ultimate godlike perfect paragon of
moderation and
excellent judgment, come down somewhere in the middle..
> The Cyc corpus, on the other hand, while useless (no point in mincing
> words) for achieving general intelligence, might indeed be valuable as a
> dictionary for a generally intelligent AI that was just beginning to
> embark on the project of acquiring the English lexicon and an
> understanding of the human-world domain. In fact, one of the reasons why
> I regard OpenMind/GAC as pointless is that the Cyc corpus would provide a
> superior substitute for any possible use.
yeah, but Cyc doesn't contain all the same kinds of knowledge that OpenMind
type
system does. For instance, it contains very little uncertain knowledge,
with degree
of uncertainty even roughly qualified... OpenMind is full of this...
ben
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT